Neuen et al [19] reported the meta-analysis data of EMPA-REG Outcome, CANVAS Program, CREDENCE, and DECLARE-TIMI 58. This meta-analysis demonstrated that SGLT2 inhibitors reduced the risk of ESRD and death due to kidney disease (relative risk of 0.67). This beneficial effect has been shown consistently across baseline eGFR levels in those studies. However, a meta-analysis result excluding the CREDENCE trial showed that the reno-protective effects were attenuated in patients with more advanced renal dysfunction at a baseline eGFR of less than 60 mL/min per 1.73 m² [20]. The inclusion and exclusion criteria of the original randomized clinical trials were not the same. For example, the EMPA-REG trial [10] included subjects with their eGFRs of at least 30 mL/min per 1.73 m², in contrast the DECLARE-TIMI 58 study [11] included subjects whose eGFRs were equal to or higher than 60 mL/min per 1.73 m². In this regard, 25.5% of subjects in the EMPA-REG study and only 7.4% of participants in DECLARE-TIMI 58 had eGFRs that were less than 60 mL/min per 1.73 m². Furthermore, a large number of participants were not followed up, and Asian patients made up the minority of participants in each study, including 21.6% in the EMPA-REG study and 13.4% in the DECLARE-TIMI 58 study. For this reason, uncertainty remains about the reno-protective effect of SGLT2 inhibitor in Asian populations with reduced renal function. In this study, we reviewed current evidence considering patients’ baseline renal function and performed a meta-analysis to determine whether SGLT2 inhibitor can be recommended to Korean subjects with T2DM.

We searched PubMed, Embase, and the Cochrane Central Register of Controlled Library up to March 2020, using search terms including “dapagliflozin,” “empagliflozin,” “ipragliflozin,” and “ertugliflozin,” which are drugs available in Korea. We included long-term large-scale randomized placebo controlled trials whose treatment duration was at least 52 weeks in T2DM patients with more than 100 participants in total. No studies on ipragliflozin and ertugliflozin met these criteria. Finally, we included 12 articles for 11 clinical studies (Supplementary Table 1 and Supplementary Fig. 1, available online) [9,10,21,-30]. In this meta-analysis, the final eGFR or delta eGFR from baseline was compared between the SGLT2-inhibitor treatment group and placebo-control according to the method presented in the original studies, and we analyzed the data of low and high doses of empagliflozin separately.

First, we compared the eGFR levels at each time point between the SGLT2 inhibitor and control. As shown in Fig. 1A, SGLT2 inhibitor treatment showed a lower eGFR level than the control at 12 weeks (-1.81 mL/min per 1.73 m²; 95% confidence interval [CI] -3.03 to -0.58) and at 24 weeks (-1.33 mL/min per 1.73 m²; 95% CI, -2.52 to -0.15). However, a favorable effect on eGFR was observed after at least 156 weeks of treatment. At 208 weeks of treatment, the mean difference in eGFR was 3.96 mL/min per 1.73 m² (95% CI, 3.13-4.80) between groups. In general, the improving trend of eGFR was observed in the SGLT2 inhibitor group as the treatment duration was prolonged. In terms of eGFR change from baseline, the mean difference was 1.42 mL/min per 1.73 m² (95% CI, 0.42 to 2.41) at 156 weeks (Fig. 1B). Therefore, SGLT2 inhibitor treatment showed reno-protective effects after long-term treatment.

The glucose-lowering effect of SGLT2 inhibitors was consistently observed: the mean difference in the decline of the glycosylated hemoglobin (HbA1c) was -0.54 (95% CI, -0.67 to -0.41) at 52 weeks and -0.62 (95% CI, -0.75 to -0.48) at 104 weeks, respectively (Fig. 2A). The magnitude of body weight reduction was also greater in the SGLT2 inhibitor treatment than control treatment group (Fig. 2B). Both the systolic and diastolic blood pressure were more significantly decreased in the SGLT2 inhibitor treatment (Fig. 2C, D).

There was no difference between the SGLT2 inhibitor and control groups in hypoglycemia events and urinary tract infection (Fig. 3A, B). However, more subjects were diagnosed with genital infections (risk ratio [RR], 3.34) (Fig. 3C) and diabetic ketoacidosis (RR, 2.22) (Fig. 3D). Acute kidney injury was less in the SGLT2 inhibitor group than in the control group (RR, 0.71) (Fig. 3E), but volume depletion was slightly more common in the SGLT2 inhibitor group (RR, 1.16) (Fig. 3F).

The clinical studies we included for this systemic review and meta-analysis did not enroll a sufficient number of patients with eGFRs of less than 60 mL/min per 1.73 m². The patients with eGFRs of less than 60 mL/min per 1.73 m² included 7.4% of participants in the dapagliflozin study (DECLARE-TIMI 58) and 25.5% of participants in the empagliflozin study (EMPA-REG). Dapagliflozin did not demonstrate the prevention of eGFR decline in patients with eGFRs of less than 60 mL/min per 1.73 m² compared to the placebo group during four years of follow-up (P = 0.053) [11]. This negative result for preventing eGFR decline with dapagliflozin could be due to the insufficient number of enrolled patients with eGFRs of less than 60 mL/min per 1.73 m² since dapagliflozin reduced the rate of eGFR decline in patients with eGFRs of more than 60 mL/min per 1.73 m². In the empagliflozin study, patients with eGFRs of less than 60 mL/min per 1.73 m² treated with empagliflozin showed higher eGFRs compared to the placebo group from 156 weeks, while the analysis with total enrolled patients showed higher eGFRs at 104 weeks (Fig. 4A vs. Fig. 1A). In terms of the eGFR change from baseline, SGLT2 inhibitors showed a beneficial effect at 208-week treatment compared to the control group (Fig. 4B). Both SGLT2 inhibitor studies revealed the initial decrease of the eGFR after SGLT2 inhibitors treatment since the mechanism of SGLT2 inhibitors may be due to relieving vasodilation of the afferent arteriole and following decrease of glomerular hyperfiltration in diabetes [31]. This initial decrease of the eGFR after SGLT2 inhibitor treatment was more significant in patients with eGFRs of less than 60 mL/min per 1.73 m².

The risk of a sustained decrease in eGFR by at least 40% to less than 60 mL/min per 1.73 m², ESRD, or renal death was lower in the dapagliflozin group than those in the placebo group, but there was no statistical significance in patients with eGFRs of less than 60 mL/min per 1.73 m² (hazard ratio, 0.60; 95% CI, 0.35 to 1.02; P = 0.059) [11]. In the EMPA-REG study, incident or worsening nephropathy was significantly lower in the empagliflozin group with eGFRs of less than 60 mL/min per 1.73 m² than in the placebo group (hazard ratio, 0.58; 95% CI, 0.47 to 0.71; P < 0.001) [10]. The adverse events in the EMPA-REG study were similar between the empagliflozin group and the placebo group in patients with eGFRs of 60 mL/min per 1.73 m² or more and in patients with eGFRs of 59 mL/min per 1.73 m² or less. In our meta-analysis, patients with eGFRs of less than 60 mL/min per 1.73 m² showed less hypoglycemia and more genital infections in SGLT2 inhibitor treatment compared to the control (Supplementary Fig. 2).

We arbitrarily defined Asian-dominant when Asians made up more than 40% of the total subjects. Kaku et al [21] analyzed the subgroup data of the Asian population from the EMPA-REG OUTCOME trial, which was the longest and largest trial included in our meta-analysis. Asian-dominant studies showed no significant difference in eGFR change between groups (Fig. 5A). Only one study was analyzed after 156 weeks of treatment in this meta-analysis; therefore, it is not conclusive as to whether Asians may experience similar benefits of renal preservation with long-term treatment of SGLT2 inhibitors compared to the non-Asian population. Other parameters, such as HbA1c, body weight, blood pressure, and adverse events were similar with meta-analysis of the total population (Fig. 5B and Supplementary Fig. 3).